Content uploaded by Nursuriati Jamil
Author content
All content in this area was uploaded by Nursuriati Jamil on Dec 02, 2014
Content may be subject to copyright.
A preview of the PDF is not available
Automated Grading of Palm Oil Fresh Fruit Bunches (FFB) Using Neuro-fuzzy Technique
Abstract and Figures
Automated fruit grading in local fruit industries are gradually receiving attention as the use of technology in upgrading the quality of food products are now acknowledged. In this paper, outer surface colors of palm oil fresh fruit bunches (FFB) are analyzed to automatically grade the fruits into over ripe, ripe and unripe. We compared two methods of color grading: 1) using RGB digital numbers and 2) colors classifications trained using a supervised learning Hebb technique and graded using fuzzy logic. A total of 90 images are used as the training images and 45 images are tested in the grading process. Overall, automated grading using RGB digital numbers produced an average of 49% success rate, while the neuro-fuzzy approach achieved an accuracy level of 73.3%.
Figures - uploaded by Nursuriati Jamil
Author content
All figure content in this area was uploaded by Nursuriati Jamil
Content may be subject to copyright.
... Many methods have been introduced to address the obstacle and implement deep learning techniques to categorize the ripeness of oil palm fruit. Jamil et al. [22] established the first artificial intelligence system for oil palm fruit ripeness classification in 2009. Their AI system uses a Neuro-Fuzzy model that had been trained on color data collected from 90 images. ...
An accurate image retrieval technique is required due to the rapidly increasing number of images. It is important to implement image annotation techniques that are fast, simple, and, most importantly, automatically annotate. Image annotation has recently received much attention due to the massive rise in image data volume. Focusing on the agriculture field, this study implements automatic image annotation, namely, a repetitive annotation task technique, to classify the ripeness of oil palm fruit and recognize a variety of fruits. This approach assists farmers to enhance the classification of fruit methods and increase their production. This study proposes simple and effective models using a deep learning approach with You Only Look Once (YOLO) versions. The models were developed through transfer learning where the dataset was trained with 100 images of oil fruit palm and 400 images of a variety of fruit in RGB images. Model performance and accuracy of automatically annotating the images with 3500 fruits were examined. The results show that the annotation technique successfully annotated a large number of images accurately. The mAP result achieved for oil palm fruit was 98.7% and the variety of fruit was 99.5%.
... Inductive type was introduced by Harun et al. [16] that utilized the frequency variations to measure the maturity of oil palm fruit bunches. The sensitivity of the inductive concept was improved from the above research in [17] and a dual flat-type shape of the air coil sensor was introduced in [18]. Jamil et al. [19] and Fadilah et al. [20] utilized a neural network and the fuzzy technique to classify the ripe oil palm fruit as further development in this area. ...
Palm oil sector is considered one of main economical contributions in countries such a Malaysia where approximately 6.1% of their Gross Domestic Product (GDP) is contributed. Low Oil Extraction Rate (OER) from poor quality fresh fruit bunches have led to a decrease the oil palm production. The traditional method of using human naked eye to inspect the Fresh Fruit Bunches (FFB) during reception process is one of common methods that is still being practiced which eventually would lead to inaccuracy in grading and harvesting. This paper presents a new approach to identify the maturity of oil palm fruit using fruit-based maturity sensor and the electronic circuit. The study utilizes multiple terminals to evaluate the sensitivity of the sensor by analyzing the relation between the load voltage and the moisture content in the oil palm fruit. The sensitivity of the sensor increased by three-fold when four terminal fruit battery was used compared to single terminals with 47.61% for the moisture range of 50-80% value. This proposed approach significantly improved the accuracy in grading the oil palm fruit which is the most common challenge posed during the reception process.
... Non-optical methods usually measure the basic molecular content of all fruits such as moisture, which influences the overall performance www.nature.com/scientificreports/ of the system. The optical detection method can be further subdivided into 3 groups: imaging 5,[17][18][19][20][21][21][22][23][24][25][26] , spectroscopy 6,[27][28][29][30] and spectral imaging [31][32][33][34][35] . Among all the methods introduced earlier, imaging is the most popular method to date due to its portability, low cost, and ease of implementation. ...
The oil yield, measured in oil extraction rate per hectare in the palm oil industry, is directly affected by the ripening levels of the oil palm fresh fruit bunches at the point of harvesting. A rapid, non-invasive and reliable method in assessing the maturity level of oil palm harvests will enable harvesting at an optimum time to increase oil yield. This study shows the potential of using Raman spectroscopy to assess the ripeness level of oil palm fruitlets. By characterizing the carotene components as useful ripeness features, an automated ripeness classification model has been created using machine learning. A total of 46 oil palm fruit spectra consisting of 3 ripeness categories; under ripe, ripe, and over ripe, were analyzed in this work. The extracted features were tested with 19 classification techniques to classify the oil palm fruits into the three ripeness categories. The Raman peak averaging at 1515 cm−1 is shown to be a significant molecular fingerprint for carotene levels, which can serve as a ripeness indicator in oil palm fruits. Further signal analysis on the Raman peak reveals 4 significant sub bands found to be lycopene (ν1a), β-carotene (ν1b), lutein (ν1c) and neoxanthin (ν1d) which originate from the C=C stretching vibration of carotenoid molecules found in the peel of the oil palm fruit. The fine KNN classifier is found to provide the highest overall accuracy of 100%. The classifier employs 6 features: peak intensities of bands ν1a to ν1d and peak positions of bands ν1c and ν1d as predictors. In conclusion, the Raman spectroscopy method has the potential to provide an accurate and effective way in determining the ripeness of oil palm fresh fruits.
... For instance, Alfatni et al. [21] used the mean value of the red, green, and blue channels of palm oil FFB images as features with a simple rule-based algorithm for classification. The features were also utilized by Jamil et al. [22], but with a neuro-fuzzy algorithm for classification. Similarly, Fadilah et al. [23] used simple color features for classification, in this case, selected hue values from all pixels in the palm oil FFB image. ...
Oil palm fruit farming is one of the most leading agriculture industries in the South East Asia region. Unfortunately, most of the harvesting method is still done through manual labor. Multiple research has been conducted to help farmers automatically detect the ripeness level of the oil palm fruit. We conducted an experiment of detecting the
ripeness level of oil palm fruit by using one of the state-of-the-art
computer vision approaches, which was deep learning with a Convolutional Neural Network (CNN). The dataset used consists of 7 levels of ripeness with 400 images of oil palm fruits. The models used in this test scenario were both AlexNet and DenseNet. The result of this study showed that DenseNet has been proven to outperform AlexNet by 8.5% in terms of accuracy and 8% in F1 score.
... Kurtulmuş, F., &Ünal, H. [7] Used a CCD scanner in order to acquire rapeseed images and discriminate among their varieties 3 Jamil, N et.al, [31] To automatically grade palm oil fresh fruit bunches 4 Unay, D., & Gosselin, B. [32] Grading of apples 5 Omid, M et.al, [33] Grading of raisins 6 Font, D et.al, [37] Verification of in-line nectarine variety 7 Rocha, A et.al, [38] Automated classification of fruit and vegetables 8 Mizushima, A., & Lu, R. [39] Sorting and grading of apples ...
... In 2009, the first AI system for FFB ripeness classification was developed by Jamil et al. [11]. Their AI system used Neuro-Fuzzy model trained on colour features extracted from 90 images. ...
Indonesia is one of the biggest palm oil exporters in the world. For Indonesia to stay competitive in thepalm oil industry, the harvesting and evacuating process in its oil palm plantation need to be optimized. This research introduces machinerycalled as Smart Crane Grabber. This machinery can be used for automatic harvesting and evacuation process of oil palm fresh fruit bunch. To enable automation, Smart Crane Grabber is equipped with an Artificial Intelligencesystem for automatic ripeness sorting. TheArtificial Intelligence system developed for Smart Crane Grabber achieves 71.34% accuracy by using only 400 images as preliminary training data.
Palm oil mill industries are commonly found in Malaysia. This is due to
Malaysia’s position as the second leading producer of palm oil worldwide. In
2021, Malaysia supplied a third of the world’s palm oil next only to Indonesia.
Recently, there is a trend of increasing price of palm oil in the market, marking
higher demand and higher profitability for the palm oil industry players,
including the palm oil mill operators. Towards supporting this greater demand,
there are calls for action for palm oil mills to upgrade their “stone age
technologies” in order to remain competitive. Sustainability of the operations
for future palm oil mills need to be addressed now in line with the drive to
adopt new IR4.0 technologies. IR4.0 adaptation in the palm oil mill will create
new synergy in production, operation and maintenance management
perspective. This paper reviews the current processes in a typical palm oil mill
and followed by the potential adaptations of IR4.0 in the existing processes.
Five main pillars of IR4.0, namely IoT, big data, cloud computing, machine
learning and artificial learning will be discussed in relation to smart palm oil
mills of the future. The expected outcomes include increase in the overall
production effectiveness, as well as improvement in the administration and
maintenance management of smart palm oil mills in the country.
Plant Pathology is the scientific study of plant diseases, caused by pathogens and environmental conditions (physiological factors). Detection and grading of plant diseases by machine vision is an essential research topic as it may prove useful in monitoring large fields of crops. This can be of great benefit to those users, who have little or no information about the crop they are growing. Also, in some developing countries, farmers may have to go long distances to contact experts to dig up information which is expensive and time consuming. Therefore, looking for a fast, automatic, less expensive, and accurate method to detect plant diseases is of great realistic significance. Such an efficient system can be modeled by integrating the various tools/techniques of information and communication technology (ICT) in agriculture. The objective of the present chapter is to model an intelligent decision support system for detection and grading of plant diseases which encompasses image processing techniques and soft computing/machine learning techniques.
Oil palm is one of the leading agricultural industries, especially in the South East Asian region. However, oil palm fruit ripeness classification based on computer vision has not gained many satisfactory results. Therefore, most of the ripeness sorting processes are still done manually by labor works. The objective of this research is to develop a model using a residual-based attention mechanism that could recognize the small detail differences between images. Thus, the model could classify oil palm fruit ripeness better. The dataset consists of 400 images with seven levels of ripeness. Since the number of images in the dataset, Ten Crop preprocessing is utilized to augment the data. The experiment showed that the proposed model ResAtt DenseNet model, which uses residual visual attention could improve the F1 Score by 1.1% compared to the highest F1 Score from other models in the experiment of this study.
Papaya is the largest contributor to fruit export of Malaysia. Before exportation, these fruits for export are graded according to their size, maturity and defects. The major parameter which is used to classify fruit size is their weight. Currently, the papayas are being weight individually and such practice is time consuming and labor-intensive. The advent of computers and machine vision technology offers great potential to automate this process. Therefore, the objective of this research is to estimate the papaya weight using results of the image analysis and then classify them according to their grades. The methodology involves measuring the actual volume and weight of papaya samples, and capturing their images. The characterization results showed that the weight and volume parameters are highly correlated and therefore, the derived formulation based on the collected data could be used to estimate the size of papaya. In the image processing task, the morphological procedures and segmentation using excess green color filter allow papaya images to be precisely distinguished from the background and shadow. This in turn allows the computation of the estimated volume of papaya simply by measuring the radius of the object at specific area and integrating over the length. Finally, papaya weights are estimated using the volume information. The classification ability of the proposed system when tested yields above 90% accuracy.
The oil content of the flesh of mesocarp has direct relationship with color bands red, green and blue. By running intensive experiments, it was found that oil content correlated with the red color band, with a regression value of 0.86. The finding of this study may be useful for determining the ripeness of oil palm for harvesting and for the use in the operation and control of continuous steriliser in palm oil mill.
Agriculture products are being more demanding in market today. To increase its productivity, automation to produce these products will be very helpful. The purpose of this work is to measure and determine the ripeness and quality of watermelon. The textures on watermelon skin will be captured using digital camera. These images will be filtered using image processing technique. All these information gathered will be trained using ANN to determine the watermelon ripeness accuracy. Initial results showed that the best model has produced percentage accuracy of 86.51%, when measured at 32 hidden units with a balanced percentage rate of training dataset.
Mango is one of the popular fruits produced in Malaysia and can be graded into different groups according to their size, maturity, disease and defects. Size is one major parameter that consumer identifies with the quality of fruits. Manually grading size of mango is laborious, tedious and costly. An automatic grading system like a computer vision system can be used to improve the quality of grading and reduce dependency on available manpower. Image analysis is the core of the computer vision system to achieve the required measurement and grading of fruits. The objective of this study is to develop an algorithm for grading size of mango based on the image analysis process. Size of mango can be estimated in different ways using image analysis. One method is to plot the area of mango measured by image analysis against the actual weight of mango in a graph. If correlation between the two measurements is high then it is possible to create a formula from the graph for estimating the size or weight of mango. In this study, the results show that the area measured by image analysis has high correlation with the actual weight of mango with R2 = 0.934. The results of grading size of mango using a formula created in the experiment are also satisfying and encouraging.
This study was aimed at investigating the relationship between oil content in oil palm fruit and its surface color distribution. A total of 80 fruit samples were randomly collected from the field; each sample consisted of two individual fruits. Fruit samples were photographed digitally under room temperature and controlled lighting and then subjected to total oil analysis using soxhlet extraction techniques. The digital images were first rectified using Adobe Photoshop®. Rectified images were clustered and subjected to unsupervised classification using MultiSpec Application.® Quantification of surface color distribution was performed in Arc View®. Relationship between oil content and color distribution was determined using multiple linear regression. Results showed that total oil content ranged between 35.2 and 86.4%. Significant correlation was found between total oil and all color components, with black yielding the strongest correlation (r = -0.85), followed by red (r = 0.81), orange (r = 0.62) and yellow (r = 0.48). The relationship between total oil content and color components was best explained with the following regression models: (1) %Total oil = 88.08-0.52 (%Black)+1.30 log (%Yellow), and (2) %Total oil = 36.84+0.63 (%Red)+ 1.52 log (%Yellow). Both these models explained 80-81% of the variation in fruit color with 76-78% accuracy. When validated on a separate data set, these models showed 55-56% accuracy. The benefit of harvesting oil palm fruits based on the relationship between surface color distribution and total oil was estimated as USD 0.15 per tree per year.
The quality feature of an ordinary Elgaeis guineensis oil palm was quantified using a computer vision model in order to inspect and grade the oil palm fresh fruit bunches by an automated production system. The feature considered was colour, and the inspection criteria were based on the Palm Oil Research Institute of Malaysia. The relationship between oil contents and colour was explored in HSI (Hue, Saturation and Intensity) domain for ripeness determination. Image analysis using Wilk's A and discrimination analyses were developed to inspect oil palm by four major classes: the unripe, the underripe, the ripe and the overripe. Over 400 samples were inspected from which the vision system was able to correctly classify oil palm at a greater than 90% success rate. Colour analysis was adversely affected by oil palms whose discriminant scores were located near the discrimination boundaries, and this contributed to the misclassification error. However, the misclassfication rates of vision system were consistently lower compared to human inspectors, implying that the inspection system developed has a great potential to assist humans for automated oil palm grading.
The article gives an overview of areas where computers and electronics have made a particular impact on the postharvest industry. These include environmental control and storage, quality monitoring, quality management, grading systems, inventory control, and management of product. Specific examples of the application of electronic systems have been given to illustrate the current state of postharvest technology, and some future predictions have been given. It is likely that consumer demand for improved quality, longer storage life, and guaranteed product safety will continue to grow. In a highly competitive market the industry will need to meet these demands, and electronic technology will play an increasingly important role. Improved sensors to assess quality are still needed, and handling and storage systems are likely to become increasingly sophisticated. In the latter half of the twentieth century technology has contributed much to improve the world's food supply, but it has also generated problems for the wider society, which will require attention in the next millennium.
Software for detecting the quality features of golden delicious starfruits of Averrhoa carambola L. genus were developed for automated inspection system using machine vision technology. The features considered were colour and shape. The use of artificial classifiers such as linear discrimination analysis and multi-layer perceptron neural network as a tool to detect starfruit maturities such as unripe, underripe, ripe and overripe in HSI colour space were investigated. The colour spectra of matured and unmatured fruits were characterised using all colour features ranging from hue 10 to hue 74, and, using principal hues generated by Wilks-lambda analysis. Experiments performed on 200 independent starfruit samples revealed that linear discriminant analysis after Wilks-lambda analysis was more precise in classification than direct application of linear discriminant analysis. However, the classification accuracy of multi-layer preceptron remained relatively the same before and after feature reduction. Overall, the average correct classification for linear discriminant analysis and multi-layer preceptron was 95.3% and 90.5% respectively during testing stage. Meanwhile the use of Fourier transform was investigated for shape discrimination. This algorithm produced 100% success rate in detecting starfruits by three shape categories: well-formed, slightly deformed and seriously deformed. Both colour and shape analysis was easily affected by the variation of lighting levels, and this contributed to the major classification error.
In making physical assessment of agricultural materials and foodstuffs, images are undoubtedly the preferred method in representing concepts to human brain. Regardless of the product, from fresh fruits to prepared foods, colour and moisture content are two important attributes which are regularly being sought by food as well as agricultural engineers. Although these attributes can traditionally be assessed with wide assortment of equipment, however, they generally lack spatial resolution since measurements were often integrated over a small area. This paper focuses on image acquisition technologies which can reveal the information of interest in two dimensions using visible and non-visible band of radiation. Respectively, the technologies investigated are the machine vision system and the computerised radar tomography. The first one was applied for colour grading of oil palms and the second one was used to map the moisture content in grain. The vision system correctly classified 92% of the stationary oil palms by four grade categories, and the radar tomography accurately mapped grain anomalies at 1 GHz over 12–39% moisture range.